vllm
https://github.com/vllm-project/vllm
Python
A high-throughput and memory-efficient inference and serving engine for LLMs
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Feature][Spec Decode] Carry auxiliary state across PP stages
- [ROCm][BugFix] Trim MoE GEMM input to valid token prefix
- [perf][MoE] flashinfer fp4 shared expert fusion
- [Bug]: draft model quant config never receives packed_modules_mapping — fused projections are built unquantized
- [ROCm] Disable AITER FP8/FP4 BMM for MLA by default
- [Bug]: CUDA graph memory profiling estimation is too huge
- [Bug]: Assertion failure due to inconsistent block counts between P and D sides when tools dict key ordering differs in template serialization with tool calls enabled
- docs: Add troubleshooting note for Qwen3-VL FP8 hang on Blackwell (#46619)
- [Bug][KV Offload] --kv-offloading-size allocates 1.92x the requested host memory as unreclaimable shmem
- [Bug] v1 detokenizer: client stop strings match inside the reasoning segment, decapitating think-in-prompt models (lm-eval sends stop on every request)
- Docs
- Python not yet supported